The VHA has been an international leader in the scientific measurement and management of inpatient quality, from an early system-wide electronic medical record to the careful benchmarking of the Inpatient Evaluation Center (IPEC). Substantial expertise and infrastructure have been developed for the measurement of short- term outcomes for inpatient care. Although hospitals devote considerable energy to improving long-term outcomes for patients, there is little research linking hospitals to long-term outcomes-nor are there operational systems to provide feedback to hospitals on their results. In order to develop an approach to studying long-term outcomes after hospitalization, patients with a single condition, severe sepsis, will first be examined. Severe sepsis is a very common, acutely life-threatening condition cared for by a wide variety of physicians on both general care floors and in intensive care units throughout the VHA. Sepsis is at least as common as such recognized high priority conditions as acute myocardial infarction. There is currently little information about which patients with severe sepsis are at greatest risk for adverse long-term outcomes. Furthermore, although there are a number of highly effective therapies for severe sepsis, and hospitals vary widely in their implementation of those therapies, there is little information on the extent to which long-term outcomes after severe sepsis are affected by hospital care. The fundamental goal of this proposal is to use severe sepsis as a model condition in which to develop a process for linking acute inpatient care to long-term outcomes and to develop a strategy for identifying individual hospitals that are high performers and those in need of improvement. Using a retrospective cohort with patient-level data from IPEC linked to the Medical SAS files and other data sources for all VHA acute care hospitals for 2005-2010, the objectives of this proposal are to: (1) ascertain the timing of onset of severe sepsis during hospitalizations, and refine the automated definition of severe sepsis; (2) identify patient-level factors associated with an elevated risk for adverse long-term outcomes; and (3) measure the extent of variation between hospitals in risk-adjusted long-term outcomes after severe sepsis. This project will be performed as a retrospective cohort study using existing data on all Veterans hospitalized with severe sepsis in the United States since 2005. To meet the first objective, data from IPEC will be combined with structured medical record review to define the measurement characteristics of the administrative data. As an exploratory extension, this data will be merged with free text searching of microbiology results and detailed laboratory values, to develop an improved case definition of severe sepsis for use in later Aims and future implementation research and quality improvement activity. For the second and third objectives, these data will be further augmented by VHA administrative data which will provide data on 1- year long-term outcomes: mortality, healthcare resource utilization, and nursing home use. For a subset of patients, self-reported health status will be included, via a link to the Survey of Healthcare Experience of Patients. Multilevel modeling will be performed to determine the heterogeneity between patients' long-term outcomes, and the extent to which such heterogeneity is attributable to the site to which they received care. In the future, this proposal will lead to risk-adjusted measures of long-term outcomes for severe sepsis that can be used to: (a) provide feedback to hospitals and identify best practices to improve the care of Veterans with severe sepsis; (b) improve prognostication and risk stratification of Veterans for long-term outcomes; and (c) serve as a model condition in understanding and improving the link between acute care and long-term outcomes for inpatient hospitalizations more generally.